Evolutionary methods for generating hierarchical and multi-label classifiers
The objective of the visiting project is to investigate and develop bio-inspired strategies to deal with Hierarchical Multi-label Classification (HMC) problems. Different from conventional classification problems, where there are no hierarchical relationships between classes, in hierarchical classification the classes can be subclasses or super-classes of other classes.
Among the hierarchical problems, mainly in the field of bioinformatics and text classification, there are a great number of problems in which an instance can be associated to two or more classes. These problems are known in the machine learning literature as HMC problems, and the development of efficient strategies to deal with them is a current and relevant research topic.